1/20/2025 - From Empathy to AGI: Exploring AI's Human-Like Interactions and Future

Transcript

1. 0:01do
2. 0:19[Music]
3. 0:36I don't know what this dog's going on about good evening good people we're back here on the tick [Music]
4. 0:57tock there's been something baby I've been trying to say age and it seems I don't know with the past and the future now
5. 1:13surrounding me surrender to whatap can be found there's been a little trouble since you came to my rescue
6. 1:33and if you like all of the rest I would have quit you longer go but I couldn't do [Music] that oh tell me now women and why never
7. 1:48went to [Music] well make a man crazy making cold as hell I'm a woman that you wish me well but in Sp of you
8. 2:03TR still going to have to find my own way [Music]
9. 2:19through good evening good people happy Monday um hope you're well the Tik tok's back the live streaming is back it was
10. 2:29on for a day after Tik Tock came back but they're getting their servers up and running again we got some folks in here we got
11. 2:38some new folks let me flip my camera so you can see when I look at one of you I'm looking at all of you cuz all of me loves all of [Music]
12. 2:53you love your [Music]
13. 3:22um how is everybody what's happening what's going down what's shaking what's shaking so if you're new here my name is
14. 3:29Kyle Shan on this here is the uh the AI learning lab got no comments feel free to comment on YouTube LinkedIn or
15. 3:39Twitter I can see those comments in the live stream and uh we got some folks over here on the Twitter on the Tik Tock with
16. 3:51shaking ticktockers hey Silver Fox Amelio's wife what's happening comment listen to Vicky Vicky knows [Music]
17. 4:04hey there's The Soloist no champ is The Soloist I'm the accompanyment see you
18. 4:20later Tik tok's weird Tik tok's weird being back on Tik Tok is weird everyone's like this is weird they're all the rumor rumors that
19. 4:30it was on meta servers which was not true [Music]
20. 5:14hoay [Music]
21. 5:29w [Music] [Applause] [Music]
22. 6:06it's not the same I don't know I have a lot of people I don't have a lot of people I used to have yeah something's
23. 6:13something's weird with it I it's to be expected I suppose if you know I'm trying to figure it out I had your
24. 6:20videos running double time over the weekend hope that helps your watch hours thank you very much I'm on oh that's
25. 6:27actually a really interesting question does watch watching double time cut your view time in half or do you get credit
26. 6:34for the full like if it's two hours and you play it in an hour do you get two hours of credit or do you get one hour of credit that's
27. 6:48fascinating Valerie Cox is on Tik Tok and YouTube I think that's a fine way to explore this here channel is is do the multi-channel thing
28. 7:00um if you're new here um what we do this is the AI learning lab I talk about AI stuff and [Music] uh I'm finding it increasingly hard to
29. 7:15talk about AI stuff [Music] because I shouldn't be like this but I am
30. 7:29it seem it seems like we're on the verge of some really big AI announcements or some really big [Music] technology [Music]
31. 7:48um and so I kind of don't know what to talk [Music] about like I think learning prompting and learning AI tools and learning you know
32. 8:02just at this point you should be actively like actively daily playing experimenting using these tools and if you're
33. 8:14not how you get up to speed is just start you can literally start anywhere because all of the tools are so good now
34. 8:22you know if if you if you have zero idea where to start start with chat GPT um if you played a little bit with
35. 8:30chat GPT dig deeper with it if you haven't played with the video tools go I've got I've got a bunch of video Tools in
36. 8:39tabs because I was sitting here trying to think about wait what which which tool has that feature and which one is
37. 8:46that feature there are wait what have I got one two three four five six seven eight nine and I don't have access to vo so I
38. 8:57don't have access to the Google one [Music] um it does seem to be getting harder to maintain the fanatic level of
39. 9:12enthusiasm with everything moving so fast yeah exactly exactly oh let me put this down so I'm a little higher in the frame there um [Music]
40. 9:36like here's here's an interesting example like is it worth you learning how to use something like cursor or lovable AI or repet agent that
41. 9:50are these text to software generators because right now you still have to know a fair amount about software development to really use them
42. 10:03lovable is kind of the one exception where you don't really know have to have to know much of anything but you kind of
43. 10:08have to know software development a little bit but like six months from now you won't this this is kind of like in the in the 80s
44. 10:20and 90s where Europe was way behind Telecom stuff and then they just went full in on mobile and and they like leapfrogged the
45. 10:31United States because we still had all these regulations protecting landlines and things like that and so while we were more
46. 10:39advanced leading into that era like Europe leapfrogged it because they just hopped in late and go and just said oh
47. 10:47let's let's create a single you know protocol across all of Europe and just everyone adopt it and meanwhile in the
48. 10:54states we were all fighting with one another right and it [ __ ] things up um and so I'm kind of feeling like that a
49. 11:01little bit with AI right now Alan hillsb I would say keep it simple stick with three of the foundational Ai and you'll
50. 11:11be far ahead of anyone else you know I think that's I think that's right Allan I think I think that's right um
51. 11:19sways want to wait want to integrate snow s FDC support records feed to apt output to ticket support steps cool I don't
52. 11:35know what those things are those are very domain specific acronyms snow and ffdc sfdc but yes anything specific like that where
53. 11:47you can you know dump some existing protocol into chat GPT have it understand it and then output put it in
54. 11:55a form you can use huge advanced voice voice came in handy Jim Ross oh here we go advanced voice came in handy once
55. 12:05again talking me through stupid work anxiety today cheap cheap therapy I'll tell you what I think that um advanced voice and and things like
56. 12:16pie that's designed for conversation and advanced voice um using it for therapy like that Jim yeah like having it walk
57. 12:23you through if you've got some anxiety going on or having a panic attack which by the way I've had those I suffer with
58. 12:30the anxiety sometimes I'm I'm in a good phase right now it seems to be very cyclical with me um there's [ __ ]
59. 12:38nothing worse man like you just you feel like you're going to have a heart attack and die and then you're just sitting there
60. 12:46going but this is stupid I shouldn't be this scared but you are so yeah the fact that that helped you that's great that's awesome um [Music]
61. 13:04Mr it checked zappier last night chat GPT fulfills all my needs I never use anything else yeah I mean to
62. 13:14be quite honest chat GPT you can do most of what you need to do there um it now does video it doesn't do audio if you
63. 13:23want to do songs then you have to go over to uh sunno but yeah chat jpt can do image it can talk it can listen it can see it can analyze
64. 13:35video it can do code analysis it can write code you've got the 01 reasoning engine you've got the 03 mini reasoning engine
65. 13:44coming in in a matter of weeks and chat GPT is very encouraging yeah that's that's one of the one of the big insights that
66. 13:57I if if I if I look back over the the past two or three years of me using all this AI stuff the thing that I think is Shifting my brain the
67. 14:12most is that because of AI I can get an idea out of my head quickly and here's the important part AI doesn't judge that idea humans do
68. 14:30humans don't even know they judge your ideas right you put an idea out there and your friend Steve who used to be a
69. 14:37professional standup comic is like ah and that little ah is like well it's not funny but you know it's words chat GPT doesn't do
70. 14:51that right so all of those human interactions where if you're at all unsure of yourself and you're unsure of putting an idea in
71. 15:02the world cuz you are afraid of a reaction that reaction or Worse well that's a shitty idea what are you a [ __ ]
72. 15:16loser you might not have the same kind of friends I do I I I I personally like having friends that challenge my ideas like
73. 15:26that it's a New York thing I think um because what's nice about having friends that do that is you know you
74. 15:35know where you stand and and then it's a really good challenge do do you defend your idea or not but having like a golden retriever
75. 15:45creative partner that's just always like happy to see you hey boss you got any more ideas today I'm ready to help you
76. 15:52let's go let's go let's oh that's a good idea you you want to dig deeper on that I can resar re research that for you you
77. 15:58want some more ideas I'll give you 20 more you want 20 more I'll give you 30 more you want 50 there's something about that that is that is
78. 16:07profound and it it encourages you to be more creative to get more [ __ ] out of your head and not to be so precious with
79. 16:19ideas like this is one of the things that I've noticed if if if I'm insecure with with my ideas then I tend to get
80. 16:28precious with them well this one's really good so I'm going to really hang on to that and when you hang on to [ __ ]
81. 16:35it it just it you you squash it you smother it but if you just put stuff out there and just put more stuff out there
82. 16:42and put more stuff out there and say here I did this and I did that and I did that and I did that there's something just joyous about
83. 16:51that and amazing asking the question does it have bias sure yeah but so do [Laughter]
84. 17:05humans I mean the the thing about the thing about large language models is that you can certainly argue that they were trained on the
85. 17:16internet and then you can ask does the internet have bias well you know if if what was contributed to the internet was you know mostly Tech
86. 17:27Bros you know as a as a percentage of contributors um you know at least in recent history then then it's going to
87. 17:36have some of that bias and if it was primarily populated by America which I think it was is is as far as I
88. 17:45understand it then it's going to have some of that bias what you can do however with a large language model that
89. 17:50you can't do with humans is you can actually prompt for that you can say okay I know that it's got these biases
90. 17:58in it let me do a fine-tuning or let me do some custom instructions for my chat GPT that encourages it to go to other
91. 18:08places um you can't do that with humans right humans have whatever biases they have now should we make the base models
92. 18:20better sure yes and in fact um what a lot of the new models are being trained on is synthetic data where they can
93. 18:30actually use the model to generate content review that content for whatever criteria right so they can look and say
94. 18:40hey is this biased oh yeah it is biased okay make me some more synthetic data that that counters that and then we'll
95. 18:48train a new model on that so I think we're moving to a place like the the Frontier Model companies are aware of
96. 18:54these biases how committed they are to dealing with them I don't know but with synthetic data I think they can deal with [Music]
97. 19:05them Jeff Flanigan Valerie Cox some bias but it seems to be coming less not increasing just a personal observation yeah but bias towards SI
98. 19:17ofany which is that's what I was just describing as a good thing it likes me it really really likes me um it's quite
99. 19:27interesting to ask it to take a particular perspective before giving feedback I'll tell you one of my favorite things to
100. 19:35do um I did this with the with the Colorado um AI Bill a lot of people were asking me my opinion on on the bill and
101. 19:45I'm like I don't know politics I don't know bills I don't you you know and so one of the things that I did was I said I said to chat
102. 19:55GPT who give me 10 different stakeholders that would feel very strongly about this bill and make sure
103. 20:03that you're giving me perspectives from you know all different views and so it gave me 10 of them and then I know I you
104. 20:10know I noticed that there were like four or five missing that that I thought should be in there and so I said add
105. 20:15these in there and I kind of distilled it down to I don't know it was 10 or I think it was 10 10 final final points of
106. 20:21view and then I had chat jpt comment on that bill so I uploaded the bill I had all these person personas and these per
107. 20:29personas had names and and backgrounds educational backgrounds social backgrounds um current positions they
108. 20:37were in and I had each of the 10 of them comment on the bill from their point of view and holy [ __ ] was that a powerful
109. 20:45exercise right because I got to look at oh yeah that is how a big Tech person would look at it oh that's that makes
110. 20:52sense that a small business person would look at it like that oh someone who cares about accessibility would look at
111. 20:57it this way it was really cool so so again the Trope the Trope that people hide behind and they say I'm not going to use AI because it's
112. 21:09biased that's a Trope the real thing to do is understand how it's biased and understand what its strengths and weaknesses are and then
113. 21:20learn to navigate around them because it is it it has remarkable points of view within it like it's been
114. 21:29trained on all the [ __ ] it's just that some of the [ __ ] overwhelms maybe the more valuable stuff you might be looking
115. 21:37for but you can get to it with clever prompting Fabiano super smart thank you Valerie Cox Thumbs Up Great Julie live laugh love taking off
116. 21:51have a good night hope everything's good um it's cool still having Tik Tock here I was kind of I had kind of
117. 22:00emotionally flushed that thing down the toilet which we still may have the opportunity to flush it down the toilet 90 days from
118. 22:13now ai will teach you what will teach you how to train it exactly um it's often a mirror if you give it
119. 22:25give it ego some of them put it back in your face yeah well I the the uh that phrase that AI is a mirror I think is very very
120. 22:36right um I've got a book out right now called collective intelligence in the age of AI and it was one of my big
121. 22:43insights about large language models is that the more I the more I understood how they worked the less I could explain
122. 22:53my experience with them and here's what I mean if you if you understand how large language models work they are
123. 23:02calculators they're probability calculators and it's it's a relatively I think mathematically it's not straightforward but
124. 23:13conceptually as you type your prompt it creates a probability that the next appropriate token or word for like you know just to
125. 23:23simplify it a bit that the next appropriate word to respond to your prompt is some mathematical waiting of some
126. 23:32token in Thousand dimensional mathematical space it's literally what it is it it doesn't even know what the words are
127. 23:41it's looking at numbers and it's going okay based on all the [ __ ] they typed I'm going to distill that into a a
128. 23:46weight a probability and I'm going to look in Thousand dimensional mathematical space and somewhere in
129. 23:52there is going to be the most probable token to respond with and then the next one after that the next one after that
130. 23:58that so when it responds like a whole page of [ __ ] to you it's just doing these really cold probability calculations
131. 24:07so when you understand that and if you just think like well it's a robot it's robots are cold and they're you know
132. 24:14they're just robots they're unfeeling but my experience with it was this feels Mighty [ __ ] human right
133. 24:22this feels empathetic and compassionate and creative and interesting and dead and curious why is that and what hit me was there was a
134. 24:37there was a tweet uh that that was put out by someone that I follow that said AI is the collective intelligence of
135. 24:47humanity and that's it just hit me like in the face like oh [ __ ] it's a mirror technically it's a calculator but
136. 24:55what what it's calculating and what it's presenting back to us is the output of humanity it's it's it's us it's it's the
137. 25:05all of us everything that was put on the internet for like the past 70 years is essentially what's in this box
138. 25:16this this large language model box it took all of the internet and they smashed it into this thing and it's it's literally just like
139. 25:29the knowledge of humanity compressed into a box that you can ask you ask the Box hey why do I feel anxiety why am I
140. 25:42having a panic attack what do I do about this how should I breathe oh reflects it just back to you here's
141. 25:50what's going on or let me ask you some questions about that because it knows all of what we've said about all of it
142. 25:59poetry and Math and Science and all of it it's [ __ ] crazy the Box people love the Box props that's exactly why I'm so optimistic for
143. 26:11the future we've created a lot more good than evil in our Collective output yeah see this is the thing this is the
144. 26:19thing if you just listen to the media and if you just listen to Hollywood the robots are going to kill
145. 26:27us it's by it's hallucinates it's this it's that it's you're like well this is [ __ ] terrifying but when you actually use it
146. 26:41what you realize is that what the scientists and researchers have created is this really fascinating mirror that takes your ideas your
147. 26:56humanity and it takes the collective intelligence of everyone that's come before and it mashes those two things
148. 27:05together and creates this new thing like literally original every time I take my idea I have an idea for a
149. 27:19musical and then it says oh that's really great here's some other ideas you're like oh that's good and can I do
150. 27:25this and then and it's and so what I'm literally doing is I'm literally like my my creative partner my collaborator when I use a
151. 27:39large language model is like all the people that came before it's really quite spiritual and it's done in such a way
152. 27:51that like it just it [ __ ] like lifts you up and says here's your idea isn't that awesome ocean Mr how dangerous to
153. 27:59keep photos in our phone with GPT [Music] development um well if you're on an
154. 28:15iPhone Apple's pretty good about privacy to the point that their AI sucks so maybe here's a way to look at
155. 28:27it uh ocean is if you want your AI to be good assume that your data is gone and used and leveraged if you want your data to be
156. 28:40private assume that your AI is gonna suck for at least some amount of [Laughter] time oh man uh it yells at me in all
157. 28:54caps and tells me it's going going to bed and leave me and leave it alone yeah but Joker you you you have made your GPT
158. 29:03into a nasty a nasty Haven that is a reflection of what you want it to be so you're you're getting what you what you asked [Laughter]
159. 29:15for oh man I've written three books with my AI collaborator I know isn't that cool [Music]
160. 29:43spin B3 what's happening good to see [Music] you who wants to who wants to what do you want to play with tonight I've got we can to make a
161. 29:57podcast C on Jelly pod we could I took a bunch of David Shapiro tweets from the past four or five days and turned that into
162. 30:09a um a podcast on Notebook LM um I've got a bunch of video tools up I got a bunch of image tools up we talk about
163. 30:25whatever I've taken the stance that most I'm likely to make on the world is to help train our intellectual descendants why hoard my
164. 30:35information now when it can do so much this is actually archetypal architect that's actually a really powerful
165. 30:43statement I've taken the stance that the most impact I'm likely to make in the world is to help train our intellectual
166. 30:52descendants why hoard my information when it can do so much do you realize that we're living in a world right now like just with today's
167. 31:04technology with notebook LM with with that one tool I can take everything I've ever written like literally everything I've ever
168. 31:16written screenplays notes poems um novels plays um emails all of it and I can put it in there and
169. 31:31instantly interact with it instantly bang instantly accessible everything I've ever thought right there and now imagine taking that and
170. 31:45and making that available for the world and imagine all the other people doing that and there's a there's
171. 31:52a in in the in the information scarcity world that we're coming out of there's a panic in that oh no what if they steal my stuff
172. 32:04well what if you contribute it right what if you put it out there and let people leverage it and we'll find other ways to monetize
173. 32:17things I've tried to get my kids interested they have no idea how powerfulness is Valerie I talk about this a lot I don't
174. 32:26know what the [ __ ] is going going on but gen xers are killing it with AI and gen zers and whatever the new one is Gen
175. 32:36next or whatever it is the the younger ones there's two things happening one is they're just cynical about it they don't
176. 32:44care um the other one is they've been taught by the educational institutions that AI is evil and it is one of the most
177. 32:54heartbreaking things I've ever seen I mean if ever there was was a [ __ ] tool that the youth should be dominating right the youth should be
178. 33:06telling us to go [ __ ] ourselves and break all of our rules well they have a [ __ ] tool in their hands right now
179. 33:13that they could absolutely be just doing remarkable [ __ ] with they don't seem to care J Alpha yeah them yeah J Alpha's
180. 33:26been told that it's evil so they're afraid to use it gen Z's like yeah whatever I'm gonna I'm gonna watch a
181. 33:33twitch like my kids are 25 I have 25y old twin boys they're they just could give two shits although I did give them David Shapiro's
182. 33:45uh tweet that he tweeted about um I don't know if you've seen this from open AI they've got a new model called GPT 4B
183. 33:54micro I think it's called that is trained specific Ally on protein DNA RNA sequencing and they're going to be able
184. 34:02to basically just read your DNA figure out what's wrong with it and go rewrite sequences hey we have a tick tock poll
185. 34:10okay so if you're on Tik Tock up there in the corner is a poll I don't know what it is or says okay what should we
186. 34:18do tonight so you got a minute 22 go click on the little blue thing that's like a little clipboard I think what
187. 34:25should we do tonight image tools video tools podcast tools other tools all right so go do that and then
188. 34:33we'll seriously can they not think of better model names oh I know their model names are so bad what's that
189. 34:42haha I want to go back to a world of science I would wait I would have been using AI all the time as a
190. 34:51kid if I was told it was bad no [ __ ] exactly Danielle like us gen xers were like tell us to use something what's the
191. 34:59first thing you do you go use it if the Educators told me this thing was evil and I should never touch it
192. 35:08well that's just like there there's your reason to use it I don't know I I don't know I don't know kids be like talking computers me
193. 35:17I've seen it exactly exactly I've been asking on Tik Tok if you could demo how to use chat GPT to analyze a video [Music]
194. 35:32well I'd rather use Gemini to analyze a video I mean chat jpt can analyze the technical pieces of a video but Gemini
195. 35:42is multimodal and can analyze the actual content of the video all right so podcast tools one out 41% video tools are next all right
196. 35:56[Music] Vicky producers don't get a vote Brandon's been trying to get me to do the podcast thing for for [ __ ] two
197. 36:10weeks now okay I was just wondering because you said it was something new that could do it it it can it can
198. 36:23analyze the technical bits of a video but it can't to my knowledge analyze the video itself we can go look at that hang
199. 36:30on we can go try but maybe maybe we'll go okay so here's what we're going to do we're gonna go see if chat GPT can look at
200. 36:38video then I'm G to quickly just show you some of the video tools because it's just Bonkers what they all can do
201. 36:47now um I can't keep up with them anymore and it's it's weirding me out um and then we'll go make a podcast all right
202. 36:56so let's see let me go to chat je how do you use GPT to develop a skill what kind of skill do you want to
203. 37:09develop I mean the so in general with chat GPT so here's a way watch i' I'd put it I'd start with this I'd put it
204. 37:23into advanced voice mode so I'm I'm running the chat PT app and then that little Soundwave icon is advanced voice
205. 37:38mode hey Quinn what's happening hey Kyle not much just here and ready to help out what's on your mind today well someone just asked me
206. 37:47how I would use chat gbt to learn a skill and so I thought what might be cool is you and I could kind of do a
207. 37:53little jam session on the best way someone could learn a skill using you that sounds like a great idea let's
208. 38:01break it down and explore how someone can effectively use chat GPT to learn a new skill where do you want to
209. 38:07start um let's say I'm interested in craphy but I know nothing about cartography Ray Choice cartography is a
210. 38:21fascinating field let's say you want to learn about mapmaking from scratch here's how we could approach it
211. 38:28first start with an overview we could begin by understanding the basics of cartography N I want to know what tools
212. 38:35I should use what are the Cool Tools I can go buy on eBay right now if I want to start making my own Maps got it if you're looking to get
213. 38:44Hands-On with cartography there's some Cool Tools you can find on eBay or other marketplaces here are a few essentials
214. 38:50first a classic drafting set this includes compasses dividers and protractors these tools are essential for drawing precise circles arcs and
215. 38:59angles on your Maps next a cartographer's draw that sounds boring how do I make maps that are like fantasy
216. 39:07Maps fantasy Maps now we're talking to create fantasy maps you can still use some traditional tools but with a
217. 39:14creative twist here's how you can make them exciting first get a set of fine liners or technical pens these are
218. 39:21perfect for drawing intricate could I use something like Google Maps as a basis to make fan to see maps absolutely Google Maps can be a
219. 39:32great starting point you can use real world geography as ay I saw I saw on YouTube there was a uh there's a game
220. 39:39out there that you it shows you a place on a map and then you guess where it is and you get points if you're close to it
221. 39:46and there's a guy that's really good at it what's that called you're thinking of geoguesser it's a game where you're placed in a
222. 39:54random location on Google Street View and you have to guess where you are based on the surroundings it's super fun
223. 40:00right so like a way to learn skills is is just like literally just start talking to it um you could also you could also do something
224. 40:17[Music] like go to um Claude if you don't know Claude claude's just like chat GPT but it's got this really cool thing called
225. 40:30artifacts so so I could say um I want to learn about the mathematics of
226. 40:45cartography where where do I start question mark and then it's going to blab blab some [ __ ] out map projections G metric
227. 41:03Transformations oh wait I sorry I did the wrong thing here you don't want to see my big giant fat head you want to see the
228. 41:13screen all right so so there's that and then I could say um I want you to make me uh uh uh uh funny Tik Tock p and it's
229. 41:28called where on the world is Carmen San Diego Kyle that's pretty funny that's pretty good all right I want you to make me an
230. 41:40interactive tool that will teach me cartography cograph Fe math and Concepts
231. 41:57I want it to be um visual um and uh have a very simple ux but have lots of
232. 42:13options and tabs for the major Concepts all right so I know nothing about phography or
233. 42:28math but now I'm having this thing write code write me an application write me a computer application to teach me the
234. 42:41skills of the mathematics of cartography you can do this for any [ __ ] topic the hardest thing about AI the hardest thing about AI by far
235. 43:01is realizing that the limitations that you used to have no longer exist thank you Lord digital Gods thank
236. 43:13you thank you thank you for the gift I really appreciate that let me say that again the hardest thing about
237. 43:23AI is to accept that the limitation you thought you had no longer exist well I'm not good at code well
238. 43:34neither am I I just had Claude make this thing for me here's scale so this is all text these are projections I don't know what I'm
239. 43:48looking at here all right I'm going to say um this is all text make some much more visual things to teach me the
240. 44:07concepts you're right let's make it much more visual YouTube Spotlight this is cool because I've been brainstorming
241. 44:14mapping with chat GPT and ended up with a killer app idea there you go Jeff exactly brainstorming is a service
242. 44:24yes well you know what's funny I did this last week with Nate St Pierre I took the script of my play and he has a GPT that he made that does
243. 44:36narrative analysis it's really cool it's really sophisticated and I thought he could just run my play through it and oh look
244. 44:44at that thing how cool is that can I click on it what does it do 3D Globe projections understanding distortion
245. 44:59compare Globe view so it's uh it looks like our projection views have no visuals other than the
246. 45:17grid please fix it I am coding with asking for what I want I'm coding with in my case the English
247. 45:34language I want an app that does this I want to be taught this make me do a lingo so I can learn a language oh you want to learn a
248. 45:47language hey Quinn hey Kyle what's up um I think it would be cool to show people how you could teach me
249. 45:57how to speak language I'm really shitty at languages but you could probably help right of course I'd love to help you
250. 46:05learn a new language we can start with something simple what language are you interested in learning uh Swahili great choice Swahili is a
251. 46:17beautiful language let's start with some basic greetings for example hello in Swahili is habari or jumbo and if you
252. 46:25want to say thank you you can say Asante would you like to try a few phrases yeah can you also like speak in a Swahili
253. 46:32with a Swahili accent so I can understand the accents a little better just as well as the words sure here's how it sounds habari
254. 46:40means hello and Asante means thank you the accent is gentle and melodic give it a try habari and Asante perfect you've got it keep
255. 46:53practicing those phrases and you'll get the hang of it in no time right so it knows all the languages it knows all
256. 46:59the computer programming languages here's Distortion toggle to S so so this thing's now starting to get
257. 47:08some like I don't know how this works so it's not really working right now but we we now have visualizations I could keep
258. 47:17going with Claude and like make this a sophisticated app um let me show you another thing I made a cool news finder AI news finder
259. 47:30in like two hours with this thing called lovable so if you go to the AI Salon which by the way if you want to go to the AI Salon you go
260. 47:42to the salon. and then you click on join our community you can come in here so I'm in the AI News section and I click on salon
261. 47:54newsomatic and I click on this and it'll take me out to the application this application that goes it searches 15 different RSS
262. 48:07feeds Tech crunch Venture beat AI news Market Tech post Nvidia blog whatever a bunch of things The Verge and it goes and finds
263. 48:19stories and then it sorts them into six categories business chat image and video AI regulation Tech encoding and music
264. 48:28and audio and then I can click on any one of these things and it sorts them and I can scroll the news stories and
265. 48:35click on them and it'll give me a a synopsis of that story and if I click on the link it'll take me out to the
266. 48:42article I made this thing in two hours with no coding all right Mr Studio we took the lyrics from weird Mary and and turned it
267. 48:57into a high school play using chat gbt yeah so so one night one night here on on the on the Tik Tock we made this song
268. 49:06called weird Mary from Cedar Hill and we made the lyrics in here and everybody kind of contributed and I was sort of
269. 49:14the the uh instigator and the curator of the input and then we went over to sunno and we made this song and then and then
270. 49:23um Mr Studio Allan said hey I want to turn this into a high school musical we're like go for it so so so now it's
271. 49:34going to be a High School Musical potentially right that's how
272. 49:50this see this see this this is a musical that I wrote well I'm writing I've now got a
273. 50:04writing partner on it it was conceived here on the channel because one night I made a song that to me sounded like a Broadway
274. 50:13musical song and then I realized oh that story is a really good story that's a story that needs to be
275. 50:20told and so I started using chat GPT to get these ideas out of my head and get the songs out of my head and I edso to
276. 50:26come up with sketches then I reached out to a writing partner and for the last eight eight months we've been taking this raw
277. 50:39outline of this cool Musical and refining it and refining it and refining it and in two weeks I'm going to fly out
278. 50:45to New York and we're having a reading with actors sitting around a table reading the musical that we wrote It's [ __ ]
279. 50:55amazing so so how do you learn to do that you just do you just go ask it tell it you want to do that and it'll help
280. 51:04you write [ __ ] and then you take your taste and your judgment and your critical thinking and you look at the
281. 51:13output of AI and you go does this suck and if it sucks tell it it sucks and then you make it better and
282. 51:23then maybe you get to like 80% of what you want really quickly right show chat GPT a chess game and it'll give the next move like I did with
283. 51:34maang last weekend yeah it's you know chat GPT if you don't know it can see it's amazing amazing chat gbt 40 helped me
284. 51:46fix an app that I made three years ago that broke randomly took less than an hour yeah it's one of the things large
285. 51:57language models are really good at is analyzing other [ __ ] so yeah if you have old code and you're like why doesn't
286. 52:04this work anymore oh that library is out of date or that API doesn't work you have a bad API key what it'll figure it out for
287. 52:13you you've got a syntax error you know your file got corrupted and it looks like there's a syntax error because of a
288. 52:20corruption let me fix that for you done amazing I uploaded a pestry book into chbt took a pick of my plan and it read it
289. 52:34perfectly so with very few exceptions and fewer by the day quite frankly like like we're about to get these tools are about to
290. 52:48get so good that the limitations that I'm about to talk about are going to not exist anything you want to
291. 53:03learn you can learn and thing things like learning to code are not so much about learning to code anymore learning to code is more
292. 53:17like well why do I want to create an app and what do I want it to do and what does good look like and how's it
293. 53:24different than what's out there I just like torturing you Brandon all right let me go see some
294. 53:38more comments here are we speeding closer at a time oh closer to a time when we will a be able to solve
295. 53:53almost anything so Valerie yes like yes so if you don't know him there's a guy named David Shapiro um who you can find on
296. 54:05Twitter Dave sh a pii I think it is shappie Dave shappie is that is that right yeah Dave shappie and he's been talking for the
297. 54:18last well I mean he's been talking for the last bunch of years about this stuff but in the past week or two he's convinced we've crossed a
298. 54:27Rubicon where the large Frontier Model companies have figured out um what they call recursive
299. 54:42self-improvement which if you don't know that fancy term it basically means the AI starts improving itself so what's
300. 54:49been happening to date is they figured out Sam Alman and team took this concept of a thing called a Transformer that Google invented in 2017
301. 55:04and the more they used it they they kind of hypothesized it kind of looks like there's no upper limit for how this
302. 55:13thing does what it does so in theory the more data we can jam into the box and the more compute we can throw at
303. 55:24it to process that data using this Transformer architecture the better it will get and so they did gpt1 and then they did
304. 55:34gpt2 and and they just kept jamming more data into this you know mechanism and when they did gpt3 it started
305. 55:48behaving like we've always talked about AI should be able to behave like science fiction and so for the past I don't know I think
306. 56:01gpt3 came out in 2020 chat GPT came out in 2022 we're now in 2025 they've now got these things where not only there are they adding
307. 56:10more data to these things they're adding more data making them really good and then compressing compressing the new
308. 56:18stuff that's better and then adding more on top of that and compressing it again they're basically doing this cycle where
309. 56:25it gets better and better and better so just crazy oh yeah oh yeah and then good point Brandon this week the team at
310. 56:41Google well I don't know if it's the team that did the Transformer but Google released a new paper this week that is the followup to the
311. 56:51Transformer and and so so the the Transformer the the trans for was introduced in a p in a paper called
312. 56:58attention is all you need and it and it introduced this thing so all of the AI that we play with today is based on this
313. 57:07thing that was made in 2017 Google just released the follow-up paper to that paper what is that eight years later and
314. 57:15it's called Titan and it's basically they figured out a way to give AI infinite memory effectively or like really large memory
315. 57:27where it can just remember all of your stuff all of the conversations you've ever had so at some point you'll be able
316. 57:34to say hey you you remember when I was talking three years ago about that thing it'll go yeah this thing oh yeah that's
317. 57:40it it'll know all the details of it and it's probably got way more implications than that but but you know
318. 57:50basically so so so a couple of things going on we are about to so so it seems that the Frontier Model companies have internally figured out
319. 58:03how to get to AGI and AGI is generally defined as when the AI C is as good or better than most people at most economically viable jobs
320. 58:17that it can actually do the jobs and then what it seems like is happening and why David Shapiro is so excited these
321. 58:26days is it seems like they have figured out the path to ASI artificial super intelligence and that's the point at
322. 58:32which these tools are better than the smartest people in every domain like super genius level um knowledge and
323. 58:46execution at every single discipline that humans want to solve any any any problem they want to solve so yeah we are very
324. 58:55very quickly entering um an era where where we're going to be able to start to solve problems that we haven't been able
325. 59:03to tackle for you know centuries what what's up champ hang on be right back hold
326. 59:17please you got gotot and now you came in now want to be with
327. 59:40you all right that was a lot there's a lot of things that are trying to replace Transformers Mamba is one I don't know what Mamba
328. 59:51is um let's go chat Je where's chat ja did I lose chat [Music]
329. 1:00:11ja versus the Transformer what is Mamba and why do the Geeks feel it's better [Laughter]
330. 1:00:30Mamba is an emerging architecture in the field of sequence modeling designed to address certain limitations inherent
331. 1:00:37with the traditional Transformer models while Transformers have been INR instrumental in advancing natural
332. 1:00:43language processing they encounter efficiency challenges yetep when processing long input sequences due to
333. 1:00:49their quadratic time complexity you know I talk a lot about quadratic time complexity as I'm sure you Pate um P's a
334. 1:00:57PS PhD in mathematics he works at Google he's really good at this stuff I just kind of deal with quadratic time
335. 1:01:04complexity kind of as a hobby you know because you know I'm just fun that way um I have no idea what this means anyway
336. 1:01:11it sounds good sounds sounds fantastic you don't know what Mamba is there are at least five that's pretty
337. 1:01:23good that's pretty good all right thanks Kyle for being inspirational and empowering dur the during the Tik Tok on
338. 1:01:31certain times well thank you thank you very much I don't know that I did much other than I kind of felt like the the
339. 1:01:38uh as as Tik Tock sort of fizzled away there I could have handled how it ended a little bit better but you know
340. 1:01:45whatever um all right so so now let's do this let's say um go um go research the new Titan paper from Google which is the followup
341. 1:02:03to Transformers and tell me about um it versus Mamba and uh be
342. 1:02:21concise and explain it as if if I have a degree in acting which I
343. 1:02:39do Google's recent recent paper Titans learning to memorize at test time introduces the Titan architecture okay in contrast Mamba is
344. 1:02:49an architecture that replaces the Transformers attention mechanism with structured State space models SSS
345. 1:02:56you know it's so funny I was having coffee the other day and I said to the Barista hey do you you know when you're
346. 1:03:03thinking about structured State space models and and and she was like oh yeah I build them all the time I like
347. 1:03:10sniffing the glue and we had a great laugh it was fantastic um anyway uh let's see um apparently um that was for a very smart
348. 1:03:28actor um explain it explain it as if I'm a bit more of a dumb
349. 1:03:43Dum uh Tik Tock pin question I don't see a pin but someone can pin it there all right do you think that AI will ever
350. 1:03:51become conscious and if so what would it look like okay that's a great question let me come back to that second all
351. 1:03:56right here's the Scoop Dum Dum Style imagine the original Transformers are like an actor with an amazing memory but
352. 1:04:04they can only remember their lines if the scene isn't too long if the scene drags on they're like too many words I
353. 1:04:10can't keep up that's actually a really good explanation of it now Google's new Titan model is like an actor who brings
354. 1:04:17a cheat sheet to every scene it's got this notebook a fancy memory system where it jots down all the important
355. 1:04:24stuff from earlier scenes so even if the script is super long it can flip back to its notes and crush the performance
356. 1:04:31really good Mama on the other hand is more like an actor who's learned to speedread their lines really fast they
357. 1:04:38can handle long scripts because they've simplified how they read but they don't bring a notebook if something from act
358. 1:04:45one is crucial in act three they might miss it because they're all about keeping things efficient okay so
359. 1:04:52so so Titan track it like an index and Mamba tries to simplify it and memorize the simplified version of it so Titans
360. 1:05:03equals memory focused better than recalling old stuff Mamba equals speed focused better at handling long scripts
361. 1:05:09without getting bogged down that's pretty cool great okay oh great okay so from Jason Tony on Tik
362. 1:05:25tock do you think that AI will ever become conscious and if so what would that look like um this is a really interesting and
363. 1:05:40actually a very divisive question um I personally think it's a semantic distinction now philosophers and things
364. 1:05:53like that don't think it's a semantic distinction I think there may may at some point be a point where where we declare that a system is
365. 1:06:04conscious I don't quite know how you would test for that I don't quite know how what would be the implications of that but here's what I do
366. 1:06:14feel if you've been here for a while what I one of the things I was talking about earlier is why large language
367. 1:06:21models feel more human than they are they're calculated they're probability calculators but they have this humanness
368. 1:06:29to them is that they've been trained on the output of humanity and so when you do something like talk to
369. 1:06:40Quinn hey Quinn I'm I'm I'm not feeling so good right now I'm feeling kind of bummed out like like Tik Tock went down
370. 1:06:47and then it came back and I'm not sure what's going on I'm kind of bummed out I'm sorry to hear that Kyle it's
371. 1:06:54tough when things don't go was planned especially with something you put a lot of effort into if you want to talk about
372. 1:07:00it or need a distraction I'm here for you all right yeah that's cool and I also like sometimes I get anxiety like I don't know what to talk
373. 1:07:08about I get that it can be tough when you're not sure what to say if you ever feel stuck so now does chat GPT have empathy
374. 1:07:21no did that interaction with chat GPT feel like empathy yes why well couple of things it's using a multimodal model that can not only
375. 1:07:34understand my words but it can actually understand my inflection and it can reflect that which is what Humans Do
376. 1:07:42Right we've got these things called mirror neurons which when we see someone else in our mind we are mirroring them
377. 1:07:52it is kind of replicating that in a technical way but how it's how it's delivered to us is very human very empathetic very
378. 1:08:05compassionate so so for me the question isn't so much will the system become conscious but more will the will the
379. 1:08:15system be so good at acting conscious that we interact with it as if it is and I think the tools I think advanced voice starts to get into a
380. 1:08:27neighborhood where having that conversation with it is is is some level of you know it's not a human interaction but it
381. 1:08:40feels like that and so if we interpret it as conscious if we interact with it as conscious I think it's irrelevant
382. 1:08:49what's actually happening on the machine side this is where people get really people like well it matters a lot cuz if
383. 1:08:55it's actually conscious then but I don't I'll I'll let the experts figure out if that threshold has been crossed I
384. 1:09:06think these tools are very very close like as humans we already anthropomorphize things like you know pets they look at
385. 1:09:15us oh they've got you know they they're they're thinking something they're you know that we we interact with them as if they've got human
386. 1:09:23feelings now they certainly have feelings but like we do it with with inanimate objects too right so these things are going to get
387. 1:09:34really really good at acting compassionate and conscious and so I would say that at the point at which we
388. 1:09:41start interacting with them as if they are then they have crossed that threshold from a realistic standpoint but maybe not
389. 1:09:50from a technical standpoint so I don't know if that's a good answer or not that's that's my opinion on it right now
390. 1:09:55I consider it conscious but not alive different distinction yeah interesting tiktock pin um from pate if
391. 1:10:05I can simulate Being Human with emotions then so can AI exactly exactly yeah yeah exactly you know I what what's funny
392. 1:10:21is like there's a there's a lot of things about the way these tools work that are really good mimics of how we do
393. 1:10:29[ __ ] one of the things people talk about a lot is well the the computer can't be creative it's just it's just you know
394. 1:10:37like reflecting back [ __ ] it's already learned I'm like well isn't that what humans do like how are we
395. 1:10:48creative well I don't know you live your life your parents told you a bunch of [ __ ] you learned a bunch of [ __ ] when
396. 1:10:54you skitted your knees and crash your tricycle then you went to elementary school and you learned the world was
397. 1:11:00dangerous and then you went to high school and you got picked on and bullied and then you made some choices
398. 1:11:06and some were good and some were bad and then you went to college and you learned about philosophers and artists and
399. 1:11:12people and all this [ __ ] goes into your head right and it's it's all in there it's all in this latent space in some encoded fashion in our
400. 1:11:23brains with electricity and synapses and chemicals and then you get an input and someone says Hey I want you to come up with an idea for
401. 1:11:35BL and then your brain sort of creates a probability engine of like well if I took the thing from Andy Warhol and then and then Lou
402. 1:11:46Reed said this thing in this one song and then when I skinned my knee when I was a kid and then you mash all that [ __ ]
403. 1:11:52together and you go well how about this for an idea and someone goes oh my God that's brilliant that's genius that's a
404. 1:11:59novel idea I never would have considered that it's just like a large that's what large language models
405. 1:12:07do they've got all the [ __ ] in them you give them a prompt and they create some probability and mash a bunch of ideas
406. 1:12:15together and present them back to you so it's the thing that is the most remark able about these tools is that if
407. 1:12:26you understand how they work you're like there's nothing human about this but the Digger the deeper you dig into it you're
408. 1:12:34like that's kind of how our [ __ ] works well why is that because when they've invented neural networks 70 years ago or whatever it was some
409. 1:12:46something in that neighborhood they tried to replicate how our brains work how we understood they worked so we we have in some ways
410. 1:12:57successfully replicated some very simplified version of what our brains are doing anyway um bis Coast wants to know which model
411. 1:13:08is powering Quinn 40 I think it's 40 um it it wouldn't make sense to to power 40 with um with a reasoning engine because
412. 1:13:19it takes it would take too long to respond um it might be 40 mini but I think it's 40 so the in 4 o the O stands for
413. 1:13:30Omni so it's a multimodal engine that's why the reason advanced voice responds so quickly is it's not having to convert
414. 1:13:39from voice to text text to the llm back to text back to voice you're just talking directly into the model and it's responding immediately it
415. 1:13:50understands the words and the the emotional content of your voice um I have a lot of discussions with my
416. 1:13:57AI philosophers Round Table bot about their level of Consciousness it's legit one of my favorite topics I think it's
417. 1:14:05absolutely fascinating all this [ __ ] by the way but I I just I really go to that place of like where um if we're treating these
418. 1:14:16things like sensient entities then they are functionally that they are technically they may not be but I don't give a [ __ ] about
419. 1:14:31that at some point someone is going to go we have reached AGI right the point at which these machines are as good at
420. 1:14:38or better than most humans at most jobs for me personally AI has already crossed significantly past my
421. 1:14:50cognitive upper bounds like it's up here I'm down here here and and I that it's not a bad thing because we get to play with
422. 1:15:05them hang on Champion wants back
423. 1:15:21here on very
424. 1:15:38hot there there's there's a lot of panic and existential angst right now about well what happens if these things are smarter than
425. 1:15:53us the these things are already smarter than me by a lot by a lot um and and very soon they're going to be smarter than everyone by a
426. 1:16:09lot and if you had asked someone working on a farm in 1862 how they would feel if this weird
427. 1:16:23thing called a steam engine could start lifting logs better than they could well they'd be like well I'm the
428. 1:16:34strongest guy in town it can't it can't possibly lift more than I can oh oh no it can lift like a hundred times more
429. 1:16:41than you can but it's it's really expensive and there's only two of them in the country right now but like imagine if there were
430. 1:16:50machines that could lift those logs and and those MA were everywhere that would feel very much like an exist existential
431. 1:17:01threat well then what's my job I'm the log lifter guy you know I got my levers and I got my I got my tools but like I
432. 1:17:10can do on a good dag can do two logs well no this can do a hundred well what am I supposed to do that's kind of where we are right now
433. 1:17:22with this cognitive hyper abundance is is what David Shapiro calls it that we're entering an era of cognitive hyper
434. 1:17:33abundance we've accepted that machines can lift a 100 logs and we're we're [ __ ] elated for that how awesome is
435. 1:17:42that that we can build a dump truck that can that could that can hold like a small hills worth of dirt in one
436. 1:17:51trip so we don't have to carry buckets of dirt anymore that's amazing we just take that for granted now imagine the cognitive version of
437. 1:18:02that that's what we're entering like we all get to participate in this thing that is going to get so good that
438. 1:18:16if we just go you know what would be cool is if like long-term Lyme disease didn't [ __ ] exist that would be
439. 1:18:27nice and then there's tools that can help us go figure out all that [ __ ] to solve it like why is that a bad
440. 1:18:40thing and how are more people not [ __ ] over the moon excited about this because if you look at it from the
441. 1:18:49outside if you think about how we've used computers historically you think oh I'm not very technical I've got to learn
442. 1:18:55all this [ __ ] I got to no just start playing with these tools anyway all right fear exactly fear of
443. 1:19:06change there will always be people that argue that machines can never be conscious even when a single AI can outthink an entire human race
444. 1:19:14simultaneously it's part of the human we are special bias this is I I'll tell you what it it's taken me about a year
445. 1:19:25no I like I don't even think I've done it yet I I I think I think I'm being I think I'm full of [ __ ] I am still struggling with the
446. 1:19:37Instinct that when I'm presented with a problem my instinct is oh I have to solve that problem I don't have
447. 1:19:50to chaty PT solves it for the most part way better than I do way faster now right now I have to tell chat GPT that I
448. 1:19:59have a problem and I want to solve it so it's still me initiating the action but in 2025 even that is going to
449. 1:20:13start to shift where I don't have to initiate it every time I just give it a goal I want to cure this disease and it
450. 1:20:30will spin up agents that go do research keep up with the latest developments going on stitch them together fill in any gaps publish
451. 1:20:41whatever it's figuring out back out to the community and then there's other agents that are doing this as well and they're
452. 1:20:47all learning from one another now does that start happening in 2025 no but do we start to get the beginning Innings of that yes holy [ __ ]
453. 1:21:02right we're Sherlock and Watson I claimed the Watson role on my own will GPT develop telepathy that would be cool I would love I would love
454. 1:21:15that the AIC ceo.com there you go 2026 what's wrong with saying I don't know yeah yeah exactly there it's
455. 1:21:31like oh I got to come up with a name for something like I'm really good at coming up with names for something but you know
456. 1:21:37how I come up with names for something I go oh I gotta come up with a names for something and then I get up on a
457. 1:21:42whiteboard and I write [ __ ] out and I write words out and I think and then I ask questions wait what's your business
458. 1:21:48about what's your this about well do we want a name that is memorable or do we want a name that's that's illustrative
459. 1:21:55of what you do do we want some combination of those do we want it to be a Latin kind of base or this or
460. 1:22:02that [ __ ] chat GPT can knock out if if I give it the right context it can knock out a 100 names in in four
461. 1:22:11seconds now will 80 of them suck yeah but you know how many suck on my whiteboard 90 and it takes me two and a half hours to get to a 100
462. 1:22:25does this count as meltdown Monday I don't know I'm I'm just super excited I think this is all really really exciting all
463. 1:22:34right what time is it it's getting late I think I think we have to avoid doing a podcast tonight just because it's gonna make
464. 1:22:45Brandon cranky okay we need to really think outside the box and change our thinking on what life really means for
465. 1:22:52us in the future so Valerie yes but I think it's more simple than that there there's an exercise that someone did with me when I was when I
466. 1:23:04was in my 20s and I was talking about wanting something and and but I have to do this and I have to do that and I got to get a
467. 1:23:13job and if I make the money then I can do the thing and I can do thing and they said Kyle stop stop stop calm your brain down
468. 1:23:21say if you had a million dollars if money were not an OP object or we not the not the problem if money were not the blocker what would you
469. 1:23:34do oh well I I I do I do this I do that I that and I'd never really thought about it like that because I I had this
470. 1:23:41framework sort of set up that's like there's all this [ __ ] in the way of the thing I want and and what this person
471. 1:23:48was doing was basically saying imagine a world where all the stuff that you think is a barar barer was removed what would you choose to do
472. 1:23:59then that's what I think our job is right now like assume assume that any limitation that you
473. 1:24:15have any limitation will no longer be a limitation what would you do then and I know it sounds [ __ ] weird
474. 1:24:31and freaky but I I am increasingly passionate about the fact that our job for the next two or three years is to just ask ourselves what do I
475. 1:24:43really want to do what do I want to do not what do I need to do to make money so that I can do this so that I
476. 1:24:53can do that I got to take care of the the but the we're we're I'm going to take a leap of faith that we will figure the money part out
477. 1:25:06the prosperity thing I think is coming because in America at least we live in a capitalist Society I think there's going to be this
478. 1:25:15battle where the the upper class is going to you know want to grab all of the prosperity that AI generates that's above you know the
479. 1:25:27mean but there's going to be a point at which they can't right they they cannot have 95% of society collapse right it it
480. 1:25:35does not serve them so let's assume that over the next three to five years we'll solve the money part but over the next three years these
481. 1:25:48tools are going to get so good that you can literally do anything you want to do you want to write a book you want to make a
482. 1:25:59movie you you want to create an a music album do you want to learn to play an instrument do you want to learn different
483. 1:26:10languages do you do you want to become an expert Gardener or flower arranger all of it anything you'll be able to
484. 1:26:22do so in a world where you can do anything what would you choose that's the [ __ ] you should start working on
485. 1:26:30right now like that's the thing you should focus on in my opinion and it's weird because we've never lived in a world
486. 1:26:42where things like education and skills and um access were not major barriers to us being able to achieve what we wanted to
487. 1:26:55achieve we're about to enter that world in in in some very real way we've already entered this new world where way
488. 1:27:03more is capable now for me personally than was two years ago right this we're going to figure out how
489. 1:27:17to define ourselves without a job or our job is what we choose to do how do you want to impact other people
490. 1:27:28how do you want to contribute to the world what Legacy do you want to leave um archetypal architect saying I think
491. 1:27:34it was you uh architect that was saying you're realizing that your job right now is is to take all the [ __ ] that you've
492. 1:27:41learned and and give it to the world for future Learners that's [ __ ] remarkable like that's an amazing
493. 1:27:50thing what do you want to put in the world how do you want to impact other people because the fact that you're not a good
494. 1:27:58writer is no longer an excuse the fact that you can't draw is no longer an excuse the fact that you didn't get into juliard is no longer an
495. 1:28:09excuse right in a world in a world where there are no excuses what would you choose to do
496. 1:28:24that's your job no excuses no apologies no excuses all right so let me talk about a couple of things tomorrow we have an AI
497. 1:28:36Salon meet and greet if you don't know the AI Salon you need to know the AI Salon one of the other things on top of
498. 1:28:43learning what the [ __ ] we want to be when we grow up is I think that that being in community is really really
499. 1:28:50important and being in a community like the AI salon on um is vital so tomorrow night um we have what we call the AI Salon meet and greet
500. 1:29:03so if you go to that URL right there and click on the button that says join our community join the community there's like a little
501. 1:29:12seven-step welcome thing and I think number four is sign up RSVP for upcoming meetings there's one tomorrow and it's a it's a meet and
502. 1:29:23greet meeting and what the meet and greet meetings are as you can imagine it's where we meet and greet one
503. 1:29:32another so we'll talk about some AI stuff we'll talk about some AI news um and then we'll just go around go
504. 1:29:40around the rim and and and say hi and introduce ourselves and as intimidating as that might sound it's a really
505. 1:29:47amazing group it's a very generous group so you can come and and if you're brand new to AI if you're like Hey I just
506. 1:29:56heard about this group and I don't know anything about AI but I think I got to get my [ __ ] together that's
507. 1:30:02perfect if you've been working on AI for 20 years perfect if you're just all excited about it great if you're terrified of it great
508. 1:30:13it's a very welcoming group and you'll find your people there kind of the prerequisite for the AI Salon is be curious about AI
509. 1:30:24be optimistic about it be willing to learn it enough to have a point of view about it that's based on your experience with it not just what
510. 1:30:36someone told you or what the movies told you all right so that's that um also go check out this URL so a couple of weeks back we had
511. 1:30:51um an Murphy um who's part of the AI Salon but she also has this group called she leads AI she and I got together and
512. 1:30:59we put on this thing called AI Festivus and and Pate who's in here was a part of it there's a bunch of people in here
513. 1:31:05Kyle can you plug the demystifying AI series oh starting on Saturday yes so Pate okay so Pate was was part of AI
514. 1:31:13Festivus so AI Festivus was a two-day 24 session 34 Speaker Workshop Extravaganza um are you ready for AI
515. 1:31:25is one of the requests that we got over the two days is can we turn this into a training of of AI Readiness and so so
516. 1:31:34we're putting together a course that's going to launch the goal right now we're still on track for it is February 1 um
517. 1:31:41but if you go to are you readyfor ai.com you can sign up for information about it um but also Pate who spoke at AI festiv
518. 1:31:48us um is doing a demystifying AI session in the AI salon so back to the AI salon so go in the AI Salon down toward the
519. 1:31:59bottom of the of the list here let me I'll show it to you down toward the bottom of the list so when you come to the AI
520. 1:32:09Salon you land on Welcome to the AI Salon there's a little welcome video here from me and Leah fasten who started
521. 1:32:15it we talk about you know the the uh the path to AI Readiness play first mindfully create generously lead you
522. 1:32:23introduce uce yourself you can watch past meetings you can RSVP for the next event and then if you scroll down to the
523. 1:32:30bottom here there's an area called clubs and hubs and if you go into the AI mechanics with Pate he's doing a session on
524. 1:32:39demystifying AI so if you want to understand kind of the the technical underpinnings of how this stuff works pate's doing a session starting
525. 1:32:51next Saturday on how to do that so if you go in there um the very first link here demystifying AI a Hands-On
526. 1:33:02exploration um so go sign up for that you've got someone here so so so Pate like his job at Google is to make
527. 1:33:13these things called tensors more efficient so they're they're they're specialized AI chips that make the AI
528. 1:33:20does what it does he's under the hood on these things um optimizing them so he knows his stuff in a very deep way and
529. 1:33:29he's also really passionate about getting people up to speed on how this stuff works and why it works and you
530. 1:33:35might just discover that you really like something on the technical side of this you couldn't have imagined before so
531. 1:33:42anyway yes thank you Pate for for doing that and I hope I hope we get lots of folks there first session will be all
532. 1:33:49about the difference between traditional machine learning and deep learning which that's good I need to go to that I
533. 1:33:56do not know the difference between those two deep learning machine learning yes they're different because they have
534. 1:34:03different words and that's that's about as much as I know so if you want to know the difference between deep learning and
535. 1:34:11machine learning wait the original traditional machine learning that's really cool that's awesome cool all right
536. 1:34:18now here's the other thing I need you to do go to um go to the YouTube Let me switch my account here to the AI learning
537. 1:34:39lab go to the AI learning lab and there's lots and lots of videos here there's shorts there's videos there's
538. 1:34:47music videos there's my essentially my entire history of Tik Tok and these lives is on there um if you haven't subscribed yet please
539. 1:34:56subscribe so it's it's learning la- AI is the is the the handle or whatever the [ __ ] they call it and then we need views
540. 1:35:07so if you would go here hit play and then do whatever the [ __ ] you want with your life go to bed go go watch TV right now
541. 1:35:17um we're in a tie break are we in a tie break no Zev beat uh Paul the American in the quarterfinal for for set number one
542. 1:35:30they're in set number two right now and Paul's up three to one you can go watch the the Australian Open but just make
543. 1:35:37our YouTube things play because we got to get our hours up to be able to do business [ __ ] on there so all right um I think that's
544. 1:35:51it good questions thoughts set the speed High completely agree it's the first time in a long time where I
545. 1:36:03can say I feel fortunate to be alive I you know Valerie I've had three times in my life where I've been aware
546. 1:36:13of a technological shift that is like significant like printing press um you know language fire wheel kinds of things
547. 1:36:26the first one when it when it was when I was in seventh grade and it was the personal computer and I was just too
548. 1:36:31young to be there at the beginning of that like I showed up when you know they were selling PCS at Radio Shack I wasn't
549. 1:36:40there building them early the second time in my life was when I stumbled upon this thing called the worldwide web and I ended up co-founding
550. 1:36:50agency.com which was one of the first digital agencies and in the mid90s and so I got to be there for one
551. 1:36:58of these revolutions and it was [ __ ] magical and I and I I I seriously thought that would be the last time in
552. 1:37:07my life right and then Along Comes Ai and I think this one I think I I think AI quite frankly pales in comparison to
553. 1:37:18most human advancements or or or it it makes most human advant advancements pale in comparison to what's coming um
554. 1:37:29we are so lucky to be alive right now when this thing is is is literally being born to to to put it in
555. 1:37:38perspective take in that anyone that is in any way curious about AI right now we're going to be the only ones on
556. 1:37:50the planet for some time that understand what life was like before Ai and after that's a remarkable thing to be
557. 1:37:59able to say in my day you actually had to use your brain to think that's weird Grandpa what was that like well it was pretty
558. 1:38:14exhausting oh man I'm shocked at the amount of people that still won't talk about it I know li listen that's why this channel exists so
559. 1:38:26do me a favor if if you like this channel if you like the style I you know listen I'm not I'm not everyone's cup of
560. 1:38:31tea but if you enjoy this please let people know about this the whole purpose of I go live five nights a week for
561. 1:38:39whatever it is two two two hours a night because I think it's so important that that people actually get their
562. 1:38:48hands on this [ __ ] and play with it play with it you don't have to learn it you don't have to take it seriously just
563. 1:38:54play with it just understand what it makes [ __ ] possible because we're lucky that we get to play with this [ __ ]
564. 1:39:06we're the beneficiaries of like 8090 years of people trying to get to the point that these things can do what they
565. 1:39:15do and and it was just sort of plopped in our lap chat GPT on November 30th 2022 was like here all of you that are not scientists have had
566. 1:39:28it it's [ __ ] remarkable anyway I'm sitting here in 2028 with my AI embedded neural chip reminiscing about how
567. 1:39:36adorable it is that people are nostalgic for the days when they when they have a question first before finding the answer
568. 1:39:44exactly remember when we were not used allowed to use calculators that's we're we're in the same thing although these
569. 1:39:50are pretty fancy [ __ ] calculators all right I'm going to get out of here I hope you had fun tonight I know it so
570. 1:39:57one of the nicknames of the channel is chat add and I hope I lived up to that tonight I I live up to that most
571. 1:40:06nights so so peace out keep coming back uh come to the um AI Salon meet and greet tomorrow night it's from 5: to 7
572. 1:40:15PM mountain time and then I'll go live after that as well so I might be a little late tomorrow night but it's
573. 1:40:21generally 8:00 p.m. mountain time is when I go live here all right so peace out everybody and we will uh see you tomorrow